With thanks to Andreas Nägeli and Olivia Eliasson, Genovis, Lund, Sweden

Summary

• This application note describes a workflow to quantitatively compare intact and peptide data from trastuzumab emtansine.
• The ADC workflow in Byos® provides automated deconvolution, peak integration, and calculation of the DAR for ADCs.
• Relative levels of conjugated peptides, generated using the arginine-specific protease GingisREX® (RgpB, Genovis AB), were overlayed onto the deconvoluted intact spectra using a novel intact reconstruction algorithm.
• Intact reconstruction from peptide mapping data effectively demonstrates correlation between the two orthogonal techniques.

Introduction

Lysine-based conjugation is a widely used non-specific conjugation strategy for the development of Antibody Drug Conjugates (ADCs). Conjugation at lysine residues results in highly heterogeneous products that pose challenges to current characterization methods. These challenges include accurate drug to antibody ratio (DAR) calculation, Site specific localization and quantitation of site occupancy.

The average antibody-to-drug ratio (DAR) is used as a quantitative assessment of the distribution of drug molecules and is also as a measure of the drug potency. DAR calculation is measured at the intact level. This has been automated in the ADC workflow in Byos.

The latter two assays (Site specific localization and quantitation of site occupancy) are often determined by bottom-up MS approaches and typically use lysine specific proteases (trypsin or LysC). As lysine conjugation results in variable modification, the use of such proteases results in complex mixtures of peptides (of varying length) which can be challenging when quantifying site occupancy. A way of reducing this complexity is to utilize an arginine-specific protease to reduce the complexity of the resulting ADC peptide map.

This application note describes how GingisREX was used to develop a simplified peptide mapping protocol for a lysine conjugated ADC. Plasma stability studies were deployed to generate samples with variable DAR values and other biotransformations resulting from the loss of linker and payloads. Both intact and peptide mapping projects were created in Byos separately. Quantitative peptide mapping results were exported into the intact analysis and used to directly compare quantitative data from the two orthogonal techniques using the intact reconstruction algorithm.

Experimental

Intact analysis and DAR determination

Trastuzumab  emtansine was deglycosylated in an automated fashion using PNGaseF immobilized on magnetic beads (Genovis AB) for 30 min at 37°C in PBS. The resulting samples were analyzed at the intact level by reversed-phase LC-MS on a Waters™ BioAccord™ system equipped with a Waters™ BioResolve™ RP mAb column (2.1 x 50 mm).

Trypsin digestion of ADC samples

Samples were reduced and denatured in 5 M Gnd-HCl, 3 mM DTT for 30 min at RT, followed by alkylation with 7 mM IAA for 20 min. IAA was quenched by addition of 14 mM DTT and the reaction was diluted to 0.6 M Gnd-HCl with RapiZyme Trypsin digestion buffer (Waters). RapiZyme Trypsin (Waters) was added at a ratio of 1:5 followed by digestion for 2 h at 37°C. The reaction was stopped by addition of 0.1% acetic acid. The resulting peptides were separated on a Waters™ ACQUITY Premier CSH C18 column (2.1 x 150 mm) and analyzed by MS/MS on a Bruker Impact II QTOF mass spectrometer.

Arginine specific digestion of ADC samples using GingisREX

Samples were reduced and denatured in 7.9 M urea, 5 mM DTT for 30 min at 30°C, followed by alkylation with 10 mM IAA for 30 min. IAA was quenched by addition of 11 mM DTT and the concentration was adjusted to 1 mg/ml antibody and 6 M urea. GingisREX was added at a ratio of 1:50 followed by digestion for 4 h at 30°C. The reaction was stopped by addition of 1% formic acid. The resulting peptides were separated on a Waters™ BioResolve™ RP mAb column (2.1 x 150 mm) and analyzed by MS/MS on a Bruker Impact II QTOF mass spectrometer.

In vitro plasma stability assay

Trastuzumab emtansine was incubated for 24-48 hours at 37°C in bovine plasma at an ADC concentration of 0.2 mg/ml. The ADC was isolated by affinity capture and prepared for MS analysis according to the protocols above.

Data Processing

Raw data files from intact and peptide analyses were processed in Byos using the ADC and PTM workflows respectively. As Byos is vendor neutral software, data can be taken from both Waters and Bruker LCMS systems.

Results

Peptide Analysis

To assess the specificity of GingisREX and the ability to accurately detect and identify the resulting large digestion products. A control digest was performed using unconjugated trastuzumab. The PTM workflow in Byos was used to identify peptides. This workflow utilizes the Byonic™ search engine to identify peptide based on MS2 spectra. All arginine terminating peptides were confidently identified based on MS2 fragments (figure 1).

Figure 1. Peptide coverage map showing complete coverage of trastuzumab using GingisREX. 

Byonic was able to confidently identify even the largest peptide based on the annotated MS2 spectra shown in figure 2(a). This identification was further validated by the MS1 isotope plot (figure 2(b)). Green bars show the theoretical (averagine based) isotope distribution.

Figure 2. Byologic results: Annotated MS2 spectrum from the 17 kDa peptide (a). Observed isotope plot overlayed with the theoretical isotope plot (b). 

To highlight the challenge posed by tryptic digestion, a tryptic digest was performed on the control trastuzumab emtansine and the raw LCMS data was processed in the PTM workflow. The resulting coverage map in figure 3 shows incomplete coverage of lysine containing peptides. Furthermore, since conjugation prevents cleavage at lysine, this creates an issue for traditional peptide quantitation methods, as the corresponding unconjugated or deferentially conjugated peptides will not exist. 

Figure 3. Coverage map of trastuzumab emtansine showing tryptic peptides. 

Figure 4. Coverage map of trastuzumab emtansine showing GingisREX peptides. 

In contrast, the GingisREX digest yielded complete coverage of lysine containing peptides over all three samples (figure 4). Moreover, the digestion profile was simplified as different peptidoforms are present consisting of unconjugated, conjugated with one drug (at multiple sites) and conjugated with two drug molecules.

Relative Quantification of Conjugation

The PTM workflow in Byos contains customizable report templates for label-free peptide quantitation. Table 1 shows the relative amounts of conjugated peptide compared with the unconjugated form. For reporting purposes percentages of unconjugated peptides are hidden.

Table 1. Relative amounts (%) of conjugated peptides across the control and incubated samples.

This table was exported as a csv file and used in the intact ADC workflow to perform a reconstruction of the intact data. The reconstruction results can be seen in figures 6 and 7.

Determination of Site Occupancy Ratio

Many peptides contain several lysines (i.e., potential conjugation sites) and in most cases Byonic allowed accurate determination of the site of conjugation. In this example, three peptidoforms were confidently identified and accurately quantified (figure 5).

Figure 5. Site Occupancy Ratio Determination. All three conjugation sites were accurately identified from MS2 spectra. The DeltaMod scores (*) are shown for each conjugation site. Relative amounts of each peptidoform were calculated by normalization of XIC areas in the Byos report. 

Table 2. Relative amounts (%) of conjugated peptides across the control and incubated samples.

The Delta Mod. Score confirms the localization of the modification on the determined site. The confidence is determined by the presence of MSMS fragment ions between two possible sites. A DeltaMod score greater than 10.0 indicates a high likelihood that all modification placements are correct.

Table 2 shows the relative amounts of each peptidoform (shown in figure 5) based on normalized XIC areas, for the control (D0) and incubated samples.

Intact Analysis and DAR Calculation

Accurate deconvolution of multiply charged ions is critical for reliable quantitation of DAR and is made possible through Protein Metrics’ Parsimonious charge deconvolution algorithm¹, which is embedded within the ADC workflow of Byos.

Intact raw data files from the in vitro plasma stability assay were processed using the ADC workflow. DAR calculations were automatically generated in the report (shown in Table 3).

Table 3. Byos generated DAR values from the plasma incubated samples. The table is included in the default ADC report template (‘DAR Fraction’ tab) within Byos.

Intact Reconstruction

A common challenge in biotherapeutic characterization is the need to compare orthogonal methods in a systematic fashion. The intact workflows in Byos (including ADC) provide a way of reconstructing intact mass spectra from label free peptide quantitation data. Modifications (or in the case of ADCs, conjugations) at the amino acid level are convolved against each other to reconstruct a theoretical Intact mass spectrum. Comparing the theoretically reconstructed Intact mass spectrum against the experimental spectrum provides a way to compare peptide mapping data against Intact mass spectrometric analysis. The differences between the two spectra can then be used to compare how well each method correlates with the other. This can then be used to adjust experimental settings accordingly.

Figure 6. illustrates how the modification list from the quantitative peptide analysis can be imported into the intact part of the workflow for reconstruction.

Figure 7. shows a stacked view of the reconstructed data superimposed onto the intact spectra.

Figure 6. Inspection view from the intact ADC project showing how the reconstruction tool is implemented.  

Figure 7. Deconvoluted intact spectra of Trastuzumab emtansine from plasma samples. DAR was automatically calculated in the Byos report. Shaded blue areas show reconstruction from peptide mapping data.  

Conclusion

• The work here demonstrates two orthogonal techniques for the quantitation of lysine conjugation on trastuzumab emtansine.
• Digestion of the ADC using the arginine-specific protease GingisREX resulted in a simpler digestion profile that facilitated clearer determination of peptide quantitation and site occupancy ratio determination.
• A common challenge in the analysis of biotherapeutics, is the need to compare orthogonal methods in a systematic fashion. Intact reconstruction visually confirms that the quantitative data obtained from peptide mapping correlates to the intact analysis.

References

[1] J Proteome Res. 2018 Mar 2; 17(3): 1216–1226; doi: 10.1021/acs.jproteome.7b00839 Parsimonious Charge Deconvolution for Native Mass Spectrometry Marshall Bern, Tomislav Caval, Yong J. Kil, Wilfred Tang, Christopher Becker, Eric Carlson, Doron Kletter, K. Ilker Sen, Nicolas Galy, Dominique Hagemans, Vojtech Franc, and Albert J. R. Heck https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5838638/